The Research On The Detecting Technology Of External Thread Surface Defect Based On The Machine Vision

External threads are one of the most important tighten components, which arewidely used in machinery and auto industry. They usually work with the internalthreads, playing a role of connection, transmission and tightening. We must admit thatthe external thread surface detects affect the overall performance of the externalthreads. So we must efficiently exclude the screw threads that have defects to minimizethe possibility of security accident. However there is no mainstream detection methodwhich is fully applicable to detect the quality of screw threads both in china and abroad.To further ensure the quality of the screw thread, especially in such industries as cars,boats and aerospace, which require high precision and high quality, it’s urgentlyrequired to develop an online and non-contact system of screw threads to ensure thequality of products.Compared with the traditional detection technology, machine vision detectiontechnology possesses many advantages, such as high speed and precision, thecapability of non-contact and strong anti-interference, which enables its widespreaduse in measurement, location, defects detection of engineering domain. By combingthe technology of machine vision and image processing, this paper designs a qualityinspection system which includes image acquisition module and image processingmodule to ensure the quality of screw threads. In order to achieve the function of thissystem, the author conducts the following researches:①According to the general law of hardware selection of the detection systemwhich is based on machine vision, we finish the hardware selection. For thecharacteristics of thread parts, the author designs a movement of the actuator to acquirethe image of screw threads which are controlled by two stepper motors. Linear CCDcamera is applied to capture the surface image of the screw thread in real-time andsend the data to the computer;②As to the image processing technology, the author designs the imageprocessing method according to the characteristic of the screw threads, which includesimage smoothing, image mosaic, image enhancement, image segmentation and imageidentification. Firstly, we filter the image with proper method. Secondly, wereconstruct the screw thread image into a two-dimensional unwrapped image by usingproper image mosaic method. Thirdly, we enhance the image with the DCT to exclude the threaded texture information, and optimal threshold is obtained through thesecondary curvature. Statistical process control and morphological operations are usedto realize the defect segmentation. At last, by applying the smallest rectangle method tosummarize the defect information, the author designs a method to judge the defect.③In this paper, the author chooses the traditional camera calibration method tocalibrate the camera, using the checkerboard to calculate the pixel equivalent so thatwe can analyze the experiment quantitatively；④Analyze the experiment result and the possible errors during the experiment,and propose a method about how to minimize the error.The defects detection system which is based on machine vision can detect thedefects information of the screw thread precisely. Quantities of experiments show thatthis system can effectively detect defects located both on the crest and root of thescrew thread and realize a real time and non-contact defect, which is of great practicalvalue and social benefits.